Self-Hosting Overview

Deploy Helicone on your own infrastructure with full control over your data

Why Self-Host Helicone?

Self-hosting Helicone gives you complete control over your LLM observability platform. Run it on your own infrastructure to meet compliance requirements, keep data private, or customize the deployment to your needs.

Full Data Control

Your request logs, prompts, and analytics stay entirely within your infrastructure.

Easy Deployment

Get started with a single Docker command or deploy to Kubernetes with our Helm chart.

Open Source

Apache 2.0 licensed. Inspect, modify, and contribute to the codebase.

Production Ready

Battle-tested architecture powering thousands of production deployments.

Deployment Options

Docker (Recommended for Development)

The fastest way to get Helicone running locally. Perfect for development, testing, or small production deployments.

• All-in-One Container: Single container with all services
• Docker Compose: Multi-container setup with separate services
• Minimal Requirements: 4GB RAM, 2 CPU cores

Docker Setup

Get started with Docker in under 5 minutes →

Kubernetes (Recommended for Production)

Enterprise-grade deployment with auto-scaling, high availability, and advanced monitoring.

• Helm Chart: Production-ready Kubernetes deployment
• Horizontal Scaling: Scale services independently
• Enterprise Support: Available for production deployments

Kubernetes Setup

Deploy to Kubernetes with Helm →

What Gets Deployed?

Helicone consists of five core services:

Service	Purpose	Technology
Web	Frontend dashboard and UI	Next.js
Jawn	Backend API and request proxy	Express + TypeScript
Worker	LLM proxy logging	Cloudflare Workers (Node.js)
Database	Application data and auth	PostgreSQL
Analytics DB	Request logs and metrics	ClickHouse
Object Storage	Request/response bodies	MinIO (S3-compatible)

Architecture Deep Dive

Learn about Helicone’s architecture →

System Requirements

Minimum (Development)

• 4GB RAM
• 2 CPU cores
• 20GB storage
• Docker 20.10+

Recommended (Production)

• 16GB RAM
• 4 CPU cores
• 100GB+ storage (scales with usage)
• Kubernetes 1.24+

Quick Start

Get Helicone running locally in under 5 minutes:

1. Clone the repository

   git clone https://github.com/Helicone/helicone.git
   cd helicone/docker
   
2. Configure environment

   cp .env.example .env
   # Edit .env with your settings (optional for local dev)
   
3. Start Helicone

   ./helicone-compose.sh helicone up
   
4. Access the dashboard

   Open http://localhost:3000 in your browser.

   Create your first user account via the Supabase auth UI at: http://localhost:54323/project/default/auth/users

Available Ports

When running locally, these ports are exposed:

Port	Service	Description
3000	Web Dashboard	Frontend UI
8585	Jawn API	Backend API and LLM proxy
8787	Worker (OpenAI)	OpenAI proxy worker
8788	Worker (API)	Helicone API worker
9000	MinIO API	S3-compatible object storage
9001	MinIO Console	MinIO admin interface
8123	ClickHouse	Analytics database
5432	PostgreSQL	Application database

Next Steps

Docker Deployment

Complete Docker setup guide with docker-compose

Kubernetes Deployment

Production deployment with Helm

Architecture

Understand how services work together

Configuration

Environment variables and advanced settings

Support

Where can I get help?

• Community Support: Join our Discord for community help
• GitHub Issues: Report bugs at github.com/helicone/helicone
• Enterprise Support: Contact [e-mail] for production support and SLAs

Can I migrate from cloud to self-hosted?

Yes! Export your data using our API or MCP server, then import it into your self-hosted instance.

What's the difference between the all-in-one image and docker-compose?

• All-in-one: Single container with all services running via supervisord. Easiest to deploy but less flexible.
• Docker Compose: Separate containers for each service. Better for production, easier to scale and debug.